- Jan 22, 2013
- 27
- 0
- 10,540
Hello, everyone
Dual graphics question
I have a laptop with AMD A10-4600 CPU (7660G on-die GPU), as well as a AMD Radeon 7670 with 1 GB of GDDR3. Looking at GPU-z, I have 2 options for looking at the card(s).
1 - "7660G + 7600M", codename "Devastator" 32nm tech, 246 mm^2 die size, 1300 M transistors, 8 ROPs, 384 shaders, 5.5 GPix/s and 32.9 Gtex/s fillrates, 512 Mb DDR3 memory running at 800 MHz (RAM for the 7660D is pulled from the system RAM, and mine is 800 MHz, so that makes sense) at 128-bit and 25.6 GB/s bandwidth, and 686 MHz core clock.
2 - "7600M", which I am assuming the 7670M. Codename "Thames", 40 nm tech, 118 mm^2 die size, 716 M transistors, 16 ROPs, 480 shaders, 9.6 GPix/s and 14.4 Gtex/s fillrates, 1024 Mb of GDDR3 at 900 MHz and 128-bit bus, with 28.8 GB/s bandwidth, and 600 MHz core clock.
So, here's the question - how come I have almost double the pixel fillrate, but less than half the texture fillrate when I am running just the dedicated card compared to dual graphics? Is this because of the ROPs? Also, why does the X-Fire config show less shaders than the single? does it only show the lowest number of shaders (on the 7660G) and expect me to know that they are doubled through X-fire (hence the much higher texture fillrate)? Except that 7660G uses VLIW4 from 6900-seris , while the 7670M is still the old VLIW5 from the 5000-series.
What are some games I should run to see whether or not it makes sense to just use the 7670 and disable the onboard 7660D (I only play SC2 and Alpha Centauri on this laptop so far, with Catalyst 12.11 beta and CAP, I can play SC2 at 1080p with only minor stuttering in 4v4 battles). Seriously, GPU-z brought more questions than answers in this case.
Dual graphics question
I have a laptop with AMD A10-4600 CPU (7660G on-die GPU), as well as a AMD Radeon 7670 with 1 GB of GDDR3. Looking at GPU-z, I have 2 options for looking at the card(s).
1 - "7660G + 7600M", codename "Devastator" 32nm tech, 246 mm^2 die size, 1300 M transistors, 8 ROPs, 384 shaders, 5.5 GPix/s and 32.9 Gtex/s fillrates, 512 Mb DDR3 memory running at 800 MHz (RAM for the 7660D is pulled from the system RAM, and mine is 800 MHz, so that makes sense) at 128-bit and 25.6 GB/s bandwidth, and 686 MHz core clock.
2 - "7600M", which I am assuming the 7670M. Codename "Thames", 40 nm tech, 118 mm^2 die size, 716 M transistors, 16 ROPs, 480 shaders, 9.6 GPix/s and 14.4 Gtex/s fillrates, 1024 Mb of GDDR3 at 900 MHz and 128-bit bus, with 28.8 GB/s bandwidth, and 600 MHz core clock.
So, here's the question - how come I have almost double the pixel fillrate, but less than half the texture fillrate when I am running just the dedicated card compared to dual graphics? Is this because of the ROPs? Also, why does the X-Fire config show less shaders than the single? does it only show the lowest number of shaders (on the 7660G) and expect me to know that they are doubled through X-fire (hence the much higher texture fillrate)? Except that 7660G uses VLIW4 from 6900-seris , while the 7670M is still the old VLIW5 from the 5000-series.
What are some games I should run to see whether or not it makes sense to just use the 7670 and disable the onboard 7660D (I only play SC2 and Alpha Centauri on this laptop so far, with Catalyst 12.11 beta and CAP, I can play SC2 at 1080p with only minor stuttering in 4v4 battles). Seriously, GPU-z brought more questions than answers in this case.
. Their notebook page shows 2011 on it, therefore limited to Llano APUs
Facebook
Twitter
Instagram